another compiler better than gcj ?

[Anthony Green]

Andrew wrote:
>  In any case there's never a good cause to
> call _Jv_divI.

But of course -- I was mistaken.  This doesn't happen on x86.  I was looking
at _Jv_divI uses on another architecture, and then tested x86 with jc1 --
obviously forgetting the fact that the driver passes
down -fno-use-divide-routine for this target.  Oops!

In any case, my patch won't be in vain for other architectures. :-)

AG